]> git.deb.at Git - deb/packages.git/blobdiff - bin/parse-packages
In packages_all_$suite, store only description Id, not full description, nor
[deb/packages.git] / bin / parse-packages
index af6d21906c8c0bd7d1128a4872e9ef5f96257c30..40c4356eed0e140360366b4cb0eb29c360a0b0cd 100755 (executable)
 
 use strict;
 
+my $what = $ARGV[0] ? "non-free" : "*";
+# max. distinct results for a given package postfix
+my $MAX_PACKAGE_POSTFIXES = 100;
+
 use DB_File;
 my %packages_small = ();
+my %package_names = ();
+my %package_postfixes = ();
 my %sources_packages = ();
 my %descriptions = ();
 my @descriptions = ("we count lines one-based\000");
@@ -40,35 +46,29 @@ for my $suite (@suites) {
        tie %packages_all_db, "DB_File", "packages_all_$suite.db.new",
                O_RDWR|O_CREAT, 0666, $DB_BTREE
                or die "Error creating DB: $!";
-       open PKG, "zcat /org/ftp.debian.org/ftp/dists/$suite/non-free/{,debian-installer/}binary-*/Packages.gz|";
+       open PKG, "zcat /org/ftp.debian.org/ftp/dists/$suite/$what/{,debian-installer/}binary-*/Packages.gz|";
        while (<PKG>) {
                next if /^\s*$/;
                my $data = "";
                my %data = ();
                chomp;
                s/\n /\377/g;
-               #s/\376\377\s*\376\377/\376\377/og;
                while (/^(\S+):\s*(.*)\s*$/mg) {
                        my ($key, $value) = ($1, $2);
                        $value =~ s/\377/\n /g;
-                       $data .= "$key: $value\n";
                        $key =~ tr [A-Z] [a-z];
                        $data{$key} = $value;
                }
                # Skip double package
                next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"});
-               $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
-                       = $data;
 
+               $package_names{$data{'package'}} = 1;
                my $src = $data{'package'};
-               my $srcversion = $data{'version'};
                if ($data{'source'}) {
                        $src = $data{'source'};
-                       if ($src =~ /(\S+) \((\S+)\)/) {
-                               $src = $1;
-                               $srcversion = $2;
-                       }
+                       $src =~ s/ .*//; # strip version info
                }
+               $data{'source'} = $src;
                my $descr = $data{'description'};
                my $did = undef;
                if (exists($descriptions{$descr})) {
@@ -78,16 +78,31 @@ for my $suite (@suites) {
                        $descriptions[$did] = $descr;
                        $descriptions{$descr} = $did;
                }
+               $data{'description'} = $did;
                $packages_descriptions{"$data{'package'} $data{'version'} $data{'architecture'}"} = $did;
                $descriptions_packages{$did} .=
                        "$data{'package'} $data{'version'} $data{'architecture'}\000";
 
                my $sdescr = $descr;
                $sdescr =~ s/\n.*//s;
+               my $section = 'main';
+               my $subsection = $data{section};
+               if ($data{section} && ($data{section} =~ m=/=o)) {
+                   ($section, $subsection) = split m=/=o, $data{section}, 2;
+               }
+               $data{'section'} = $section;
+               $data{'subsection'} = $subsection;
                $packages_small{$data{'package'}} .= "$suite $data{'architecture'} ".
-                       "$data{'section'} $data{'priority'} $data{'version'} $sdescr\000";
+                       "$section $subsection $data{'priority'} $data{'version'} $sdescr\000";
                $sources_packages{$src} .=
-                       "$data{'package'} $data{'architecture'} $data{'version'}\000";
+                       "$suite $data{'package'} $data{'version'} $data{'architecture'}\000";
+               while (my ($key, $value) = each (%data)) {
+                       next if $key eq 'package' or $key eq 'version' or $key eq 'architecture'
+                               or $key eq 'maintainer';
+                       $data .= "$key: $value\n";
+               }
+               $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}
+                       = $data;
        }
 
        untie %packages_all_db;
@@ -147,6 +162,30 @@ for (my $i=1; $i<= $#descriptions; $i++) {
 close DESCR;
 untie %descriptions_db;
 
+# package names stuff:
+for my $pkg (keys %package_names) {
+       for (my $i=0;$i<length($pkg)-1;$i++) {
+               my $before = substr($pkg, 0, $i);
+               my $after = substr($pkg, $i);
+               $package_postfixes{$after} .= "$before\0";
+       }
+}
+my %package_postfixes_db;
+tie %package_postfixes_db, "DB_File", "package_postfixes.db.new",
+       O_RDWR|O_CREAT, 0666, $DB_BTREE
+       or die "Error creating DB: $!";
+while (my ($k, $v) = each(%package_postfixes)) {
+       $v =~ s/.$//s;
+       my $nr = $v;
+       $nr =~ s/[^\000]//g;
+       $nr = length($nr) + 1; # < number of hits
+       if ($nr > $MAX_PACKAGE_POSTFIXES) {
+               $v = "\001" . $nr;
+       }
+       $package_postfixes_db{$k} = $v;
+}
+untie %package_postfixes_db;
+
 rename("packages_small.db.new", "packages_small.db");
 rename("sources_packages.db.new", "sources_packages.db");
 for my $suite (@suites) {
@@ -156,3 +195,4 @@ rename("packages_descriptions.db.new", "packages_descriptions.db");
 rename("descriptions_packages.db.new", "descriptions_packages.db");
 rename("descriptions.txt.new", "descriptions.txt");
 rename("descriptions.db.new", "descriptions.db");
+rename("package_postfixes.db.new", "package_postfixes.db");